CoreNet: A Library for Training Deep Neural Networks

CoreNet is a comprehensive library for training both conventional and novel deep neural network models across a wide range of tasks such as foundation models (including CLIP and LLM), object classification, detection, and semantic segmentation. It facilitates research and engineering efforts by providing training recipes, pre-trained model weights, and efficient execution on Apple Silicon through MLX examples. The initial release includes features like OpenELM, CatLIP, and several MLX examples for improved efficiency.)

Visit Original Article →